August 2015, № 8 (183)

Shepel V.N., Akimov S.S. METHODS OF DISTINCTION SYMMETRYCAL DISTRIBUTION LAWS (FOR EXAMPLE, THE NORMAL AND THE LOGISTIC DISTRIBUTION LAW)One of the basic problems of mathematical statistics — a comprehensive study and analysis of data. This is widely known that the exhaustive characteristic of any data set is the distribution law, which governs the array. There is a fairly large number of different methods for determining the distribution law, relying solely on the data array. At the same time it should be noted that all these methods are only approximate. Also It is worth be noted that some laws of distribution have a fairly similar characteristics, making it difficult to identify them. One of these characteristics is the symmetry. Main purpose of the research — a comparison of methods of identification the symmetric distribution laws. The article considers the four methods of identification the law of probability distribution: criterion of consent by Kolmogorov, coefficient of Kurtosis, assessment Hill and the method relationships. As the laws of distribution was used logistic and normal distribution, most similar symmetrical laws. Researches performed in the laboratory of Chair of Management and Informatics in Technical Systems of Orenburg State University. For obtaining necessary data sets used by the random number generator of the program MathCad 15. To check the was generated distributions 800 (400 obeying the logistics law and 400 — normal; N = 100). For assessment of reliability the parameters of distributions remained unchanged. It was determined that the most powerful method is to criterion by Kolmogorov. Also have been studied various combinations of the proposed methods. In studying combinations shown that the most effective is the use of a pair of criterion by Kolmogorov — a method of relations. All other pairs of methods give a greater percentage of errors.Key words: logistic distribution, normal distribution, methods of distinction, a combination of methods.

Download

References:

1. Shepel V.N., Akimov .SS. Determination of the probability density of data in applied research. — Modern information technologies in science, education and practice. — Materials XI All-Russian scientific-practical conference. — Orenburg State University. — Orenburg. — 2014 pp 223-226.

2. Ayvazyan S.A., Mkhitaryan V.S. Applied statistics. Basics of Econometrics Probability Theory and Applied Statistics — M .: Unity Dana, 2007 -656 p.

3. Akimov S.S. Evaluation methods to restore law of distribution the probability and justification preferences on the basis of some of the properties of the empirical data set. — Actual problems of modern science. — 2014. — № 31. pp 139-150.

4. Akimov S.S. Methods for solving the task of restoring the probability density on a sample of the general aggregate . — Natural and mathematical sciences in the modern world: based on XIV Intern. scientific and practical. Conf. №1 (13). Novosibirsk:. "SibAK", 2014. pp 29-35.

5. Ayvazyan S.A. Distinguishing similar hypotheses about the density distribution in the form of a scheme of generalized sequential criterion. — Probability Theory and Its Applications, 1965, p. X, № 4.

6. Vapnik V.N. Restoring dependencies from empirical data. — M .: Nauka, 1979.

7. D'Agostino R. B., Pearson E. S. A further development of test departure from normality. — Biometrika, 1973, 60, №3 — p. 613-622.

8. Orlov A.I. Typical mistakes when entering the application statistics. — [Electronic resource]. — Access mode: URL: http://forum.orlovs.pp.ru/viewtopic.php?t=97 (the title of the screen).

9. 9. P.N. Dubner The calculation of direct and inverse distribution functions. // Statistics and stochastic systems, edition 15, Moscow University Press, 1971.

10. Lemeshko B. Yu, Lemeshko S.B. Comparative Analysis of criteria for deviations from the normal distribution law // Metrology. 2005. № 2. pp 3-24.

11. Bol'shev LN Smirnov NV Tables of Mathematical Statistics. M .: Nauka.1983

12. Pustyl'nik E.I. Statistical methods of analysis and processing of observations [Text] / E.I. Pustyl'nik — "Nauka ", Moscow, 1968. — 288 p.

13. A.N. Guda The models estimate the parameters of telecommunications traffic in the automated information control systems / A.N. Guda, M.A. Butakova, N.A. Moscat // Questions of modern science and practice. Univ name. V.I. Vernadsky. — 2010. — №4-6 (29). — pp 71-87.

14. Shepel V.N. Using of Hills assessment for distinguishing of the laws of probability distribution [Text] / Shepel V.N., Akimov S.S. // Bulletin of the Orenburg State University. — 2014. — № 1, January. — pp 75-78.

15. Akimov S.S. The problem of choosing the method of restoring the law of probability distribution. — International Research Journal. — 2014. — № 1-3 (20). pp 5-8.

About this article

Authors: Shepel V.N., Akimov S.S.

Year: 2015

Editor-in-chief

Sergey Aleksandrovich
MIROSHNIKOV

© Электронное периодическое издание: ВЕСТНИК ОГУ on-line (VESTNIK OSU on-line), ISSN on-line 1814-6465
Зарегистрировано в Федеральной службе по надзору в сфере связи, информационных технологий и массовых коммуникаций
Свидетельство о регистрации СМИ: Эл № ФС77-37678 от 29 сентября 2009 г.
Учредитель: Оренбургский государственный университет (ОГУ)
Главный редактор: С.А. Мирошников
Адрес редакции: 460018, г. Оренбург, проспект Победы, д. 13, к. 2335
Тел./факс: (3532)37-27-78 E-mail: vestnik@mail.osu.ru